Competition bugs+ limitations

[Ruudjah]

Introduction
I am a regular and moderator at Kurt’s Hard Goal servers (“Kurt’s”) (using energetic nick). Kurt’s is a system where players get points awarded for each game they play. Basically, the closer they end approaching the game goal, the more points they get awarded. This style has proven very addictive and Kurt’s is getting more and more popular. The system gets the best and the worst out of people, players will do _anything_ to win the game. All styles here are seen you can possibly think off: blocking opponents, building huge networks, competing on the edge at resources. People are very inventive in finding new ways to find just that little advantage over the opponent. I think rock bottom is hit where most players stumble upon some game limitations. The blocking of players is managed by having a moderator online most of the time. A few other limitations are managed by making a game play rule (which in turn are enforced by the moderators), and server-side coding.

Limitations already acknowledged
The AI contest held a few months ago, where Rondje om de Kerk won, also addressed some of the limitations/bugs in the game (I’ll call them limitations from now on, bugs sounds too negative). Rondje om the Kerk won partly by exploited these limitations very well. In the discussion about those limitations, maninthebox addressed the fact that not much people are affected by these limitations. Since of the popularity of Kurt’s, I think it is time to have a discussion about those limitations, and how to solve them. More and more people here are annoyed by the way these limitations affect gameplay. Kurt’s has almost always ~20 people online in two servers at EU times from 15:00 – 01:00 Monday/Tuesday/Wednesday/Thursday, and in weekends the servers are full around the clock. There is no doubt in my mind that Kurt’s is one of the most popular, if not _the_ most popular, OTTD gaming place. This means quite a lot of people are affected by the limitations I will describe further down this post.
Internally at Kurt’s, we already held a public discussion about these limitations and the possible solutions to them. It’s on Kurt’s forum. This link shows the discussion about the limitations, this link shows the topic about the possible solution(s). So that is the background on this topic. But read on, as most information from these two topics is integrated in this post.

Winning a game at Kurt’s is a matter of reaching the game goal before others do. The game goal can be either of two types: a NCV ranging from 3 to 17 million, or CBV ranging from 1 to 9,5 million. NCV stands for Net Company Value, where CBV stands for CityBuilder Value. NCV is OTTD’s company value corrected by some of the issues noted in the Rondje om de kerk topic. CBV essentially is the cumulative value a player created by moving passengers, mail, food and goods (water/grain/livestock/coal/wood/paper/ore etc are not calculated). It is always of utmost importance to move the most resources. This is off course logical, but I say it here to emphasize it.

Train renewal limitation
First of all, and most important, is the train renewal limitation. Imagine two players competing for the same coal mine. Several factors are present to determine the way the resources are divided. These are punctuality (rating drops as there is no train present at station), train age and having a statue. The train age factor is where the limitation in gameplay starts: in order to win the resources each resource spawn cycle, the player needs to newest train possible. Players therefore renew engines of their trains just before the train enters the station, every time, for every train. The current engine is sold, and a new one of the same type is bought. Having the engine newer then 5 weeks makes for optimum train age. Experienced and top 30 players do this for about 30-50 trains in the game, constantly renewing the train engines. You can imagine how this affects gameplay. If you have a slight less sense of imagination, I’ll suggest you try it. It is daunting. I can tell, I have a lot of experience with it. The problem furthermore lies in the fact that when one player starts doing it, you have to go along with him, otherwise you will be outcompeted and soon out of business. And people will go eventually out of business.
Now imagine two players at a coal mine competing for the produced coal. They both are very good, and replacing the engines with the fastest trains available. The system is set up in such a way, one player always wins. The winner gets about 80% of the produced resource, and the loser about 20%. If both players have all what is possible to have to improve the station ratings, the winner is determined by station building order. The last player building a station then wins. Above two limitations are for all resources, including towns.

Maximum two player at a resource
The above text also implied another limitation. This is that only two players are able to compete for a given resource, except in towns (while I believe for a certain catchment tile, this is the case also. Since you can place on different areas, more than two companies competing for passengers/mail will work). For heavy resources, such as 180+ producing coal mines, oil wells, etc, it is desirable then more than two players can compete for a given resource. Also, this should be the case for secondary resources which can produce up to 2000 tons a month such as factories and refineries. Remember: also for these resources the “80%/20% rule” applies. To give an example: a player who is just renewing trains and having bought a statue, and making sure he has always the last station, will get 80% of the goods produced of a factory, while he may have supplied all the factory input (such as timber, grain, livestock, etc).

secondary resource limitation
At secondary resources, a player moving the most primary resources (such as grain/livestock) can become In a situation where he is unable to move any of the produced secondary resources (such as food/goods). This is because two other player already may have started to compete for the goods, and bought a statue. It is impossible currently to recomplete for the goods. I think it is much more fair to base the station rating for secondary resources on the amount of primary resources brought to a factory. Hence, for a player to compete on more secondary resources, he needs to move more primary resources. That’s what the game is all about: building big networks. This rating should be based upon the primary resources brought the past 3 months. This limitation scares of new players. It happened a lot of times that a player building up a big primary resources quit after seeing a player ‘stole’ them. However we think competition is good, I do think this goes a bit too far: a player moving a lot of primary resources should gain more ‘rating rights’. This sounds a whole lot more fair or justified to me.

All limitations summed up
These four limitations are the most important limitations we encounter at Kurt’s. We think a quickfix for these limitations should be made, and included in the soonest available release of OpenTTD. In short, these limitations are:

• Newer trains win rating
• Youngest station wins
• Maximum two players at a resource
• Secondary resources rating based on primary

There are a lot more suggestions and small improvements and ideas, but the above four are really the most important limitations. These four limitations have as an effect that the gameplay is seriously negatively affected. A fix for those four limitations would really improve the gameplay a lot.

Possible solutions
The “newer trains win” limitation can be removed by simply changing the effect on the rating to 3 years. Renewing trains after three years is acceptable in gameplay. Only 3-4 renewals are needed in order to keep the rating optimal (games usually last 10-15 years). The “youngest station wins” limitation can be trivially removed too, I suppose (I read the code and it does not seem to be very hard). The “maximum two players at a resource” falls together with the removal of the 80%/20% rule. Even ratings at stations should give players even amount of resources (50%/50%). When 3 or more players compete with different ratings on a primary resource, a fair divider should be used. I am not too sure what divider. Hereby a request to discuss about a good divider for this.
The last limitation seems the hardest to remove. The rating of factories/paper factories/food factories/refineries/printing presses/sawmills based upon the input of primary resources. I think the ratings should be based on the past three months of input from a particular user. It would require a bit more coding.

[PhilSophus]

First of all, I think you would get more readers if you would structure your text a bit more, as it is quite a lot (e.g. put the paragraph before the last at the beginning and use bold-face headings for each issue).

# Newer trains win rating

From my unterstanding, there are only three stages (<1, <2 and <3 years old), so I don't know where your 5 weeks come from. As for the issue of too many replacements (or the sell train at destination "cheat" that the AI used), maybe a limit on how many vehicles can be bought per year could help (this limit could be dependent on the number of vehicles you already have). I imagine it working like a leaky bucket, e.g. you can accumulate at most Y years of the buy limit L, so that you can at most replace/buy Y*L vehicles at the same time (e.g. for replacing your fleet to a new vehicle type) and this bucket would re-fill at a rate of Y/12 per month. Such a feature could be motivated as "the vehicle producer needs to build new vehicles as it ran out stock".

Another, IMHO inferior, solution could be that buying vehicles would take time, but this would be very annoying, IMHO.

# Youngest station wins

I did a patch[1] against that some time ago. Should not be that hard to adopt to current trunk. It also fixes the issue that total cargo is reduced when two stations compete for it (that was actually the primary goal, but it also results in same amount of cargo for both, when rating is the same).

# Maximum two players at a resource

Maybe, the algorithm in the above patch might be extended to more than two competitors.

# Secondary resources rating based on primary

Seems to be related to the previous one, isn't it?


[1] Unrecognizable link in Christmas theme

[Ruudjah]

I fixed up the layout. Thanks for mentioning it.

From my unterstanding, there are only three stages (<1, <2 and <3 years old), so I don't know where your 5 weeks come from.

Me neither. But in practice, we have four stages: 5 weeks, 1 year, 2 years and 3 years. Try it

[Rubidium]

# Newer trains win rating

From my unterstanding, there are only three stages (<1, <2 and <3 years old), so I don't know where your 5 weeks come from. As for the issue of too many replacements (or the sell train at destination "cheat" that the AI used), maybe a limit on how many vehicles can be bought per year could help (this limit could be dependent on the number of vehicles you already have). I imagine it working like a leaky bucket, e.g. you can accumulate at most Y years of the buy limit L, so that you can at most replace/buy Y*L vehicles at the same time (e.g. for replacing your fleet to a new vehicle type) and this bucket would re-fill at a rate of Y/12 per month. Such a feature could be motivated as "the vehicle producer needs to build new vehicles as it ran out stock".

Or just reduce the value of the vehicle on first use (e.g. loading). Then selling the vehicle and buying a new one costs more money than the extra rating cargo and thus making it more expensive to replace the vehicles than not replacing them all the time.

[Ruudjah]

Rubidium, that is a suggestions also opted by the current number one of the list, the player named "Barcelona". He says: when selling, only get 75% of the new price. Aging devaluation then comes on top of it.

This would not help in CBV goal gaes though. CBV goal games count ALL the goods/food/pax/mail EVER moved. NCV goals, when for example building a road, loose worth. CBV do not, hence it would not be a solution for CBV goal games. Barcelona however opted this suggestion before CBV goal games hwere introduced ( about a month ago).

[PhilSophus]

Or just reduce the value of the vehicle on first use (e.g. loading). Then selling the vehicle and buying a new one costs more money than the extra rating cargo and thus making it more expensive to replace the vehicles than not replacing them all the time.

Yes, that would make sense (and by accident is also more realistic than the current behavior ). It would even make sense if the vehicle would loose something like 20%-30% already when bought, but your suggestion is more gracefully in case you have bought the wrong vehicle, so is favorable for gameplay reasons. Another sensible moment when the new-vs.-used penalty may be applied is the first time it leaves its building depot.

[barcelona]

Hello everybody,
This is my first message here
I am the same Barcelona who play on the server for Kurt.
First of all express my gratitude to all who have worked on this project.

Secondly he said that my knowledge of the English language is close to 0,

Compiled all messages in Spanish, is highly likely that the automated translation lead to misunderstandings subtract logic to my arguments.

My opinions on the current system is increasingly high, I believe that the current system must be preserved unchanged, allowing the game can be configured with the current rules,

What I mean is that. The new rules are introduced as a future option, and does not mean the automatic replacement of the current.
Or that if the substitute is allowed to configure the game optionally with the original rules.

I am going to propose some improvements in my view, always thinking of rules that are effective and easy to implement.

All the suggestions that I have to do is that the game be as realistic as possible, because I believe that any improvement in realism (subject to obvious exceptions), resulting in an improvement in the way of playing the game.

a) Download the burden on the station of origin must have the same treatment, for qualification purposes, that an accident.

b) the sale of a vehicle loaded should involve compensation for the value of the cargo not delivered and / or any inconvenience caused to passengers. In addition to assume for the purposes of rating a decline in the stations of origin.

I believe that both measures are easy to deploy and although its value would increase appears on the logic and realism of the game. To criminalize the loss of trains or some practices such as the absurd known as "truck pulls"

c) the aforementioned decline on the sale value of vehicles.

This is a matter of reality, the amortization of the cost of the vehicle has a logic accounting, but never 100% of book value of the vehicle is recoverable.
The costs associated with the purchase of the vehicle such as marketing costs, transport from the factory and putting them in depreciable, but are meaningless for the purposes of fixing the sale value of the vehicle.
In addition, the sale value of the vehicle will also be pressured by the difficulty of putting such a vehicle on the market, taking into account the rules of the game, and the crucial importance of the age of the vehicle, logic suggests that an old vehicle is very little desired and thus its value should be very low.

Perhaps a decline of 40% might be a good consideration.
(Obviously, I always refer to used vehicles, which at least has left the deport)
In addition, and also with effect from win realism, in the case of renewal of material, could be considered a small business bonus in the price of a new car, for example by 10%

I believe this measure of vital importance that penalizes the usual abuse that allows the unrealistic absolute availability and immediate delivery of equipment purchased and the very large benefit to be obtained and that allows a return on a train in just one trip. (Obviously not asking for any corrective action on these facilities, because obviously increase the realism in those terms would cause the game to be slow and boring)

d) the option to allow automatic renewal of material, at any age chosen by the player.

Currently the game and lets you change a model automatically renew for another and automatically when a model is close to the limit of its useful life. Allow to replace any age, it would be much more useful for the player and would release him from tedious routine actions.

It should be borne in mind that this does not detract strategy to the game, because if the measure c) is implemented, an abuse in the renovation of trains would be severely penalized financially.

e) In case of equal qualification sharing the same amount to all the stations draws.

f) Amendment of the rating system, taking into consideration the number of operating origins and destinations

Obviously this is far more difficult to implement, but it is also the most important and the most logical.

Obviously a greater service to society should be rewarded with greater consideration on the part of society, offering passengers 10 different destinations is much more difficult to offer a single change in the game now as easy rewards and punishes it difficult for the player who delivered a single destination (the most distant) gets all the benefits, higher revenues per load delivered and easier to obtain and maintain a good grade, however the player that gives a greater service to society is penalized, his Voting is more difficult to obtain and maintain their income and means for cargo delivered are lower.

I think the most efficient way to solve this is by establishing a counter for every population group and related goods. -25 Initialize the counter each of January and increased by 1 meter each time it is received or delivered merchandise related to a different population.

The value of the counter in every moment serves to raise or lower the rating of the assets of the group at stations that population by 1% for each point (
To identify which assets belong to two or more groups calculate the average)

If this is not well explained this suggestion f), or if you understand what I want to convey.

In addition it may be too complicated to implement, but in my opinion, if possible, improve greatly the game.

g) In any case, I believe that the sharing system should not suffer any change other than the equal distribution in case of equal qualification.

I reiterate my congratulations to all the wonderful work done by the developers, in any case my suggestions are critical to any other point of view, if not only by whether a collaboration would be of interest.

I think in general the vast majority of players, believes that the substance of the game is to build and maintain more and better communication among people, and all the proposed measures are aimed at greater efficiency in this regard. Rewarding the quantity and quality of service and punishing inefficiency and waste.

[barcelona]

# Newer trains win rating

From my unterstanding, there are only three stages (<1, <2 and <3 years old), so I don't know where your 5 weeks come from. As for the issue of too many replacements (or the sell train at destination "cheat" that the AI used), maybe a limit on how many vehicles can be bought per year could help (this limit could be dependent on the number of vehicles you already have). I imagine it working like a leaky bucket, e.g. you can accumulate at most Y years of the buy limit L, so that you can at most replace/buy Y*L vehicles at the same time (e.g. for replacing your fleet to a new vehicle type) and this bucket would re-fill at a rate of Y/12 per month. Such a feature could be motivated as "the vehicle producer needs to build new vehicles as it ran out stock".

Or just reduce the value of the vehicle on first use (e.g. loading). Then selling the vehicle and buying a new one costs more money than the extra rating cargo and thus making it more expensive to replace the vehicles than not replacing them all the time.

I have never seen such a difference for 5 months

My observation is that the computer age is by calendar year, bringing a vehicle purchased on December 30 is considered the next day as a vehicle for one year old.

[barcelona]

Rubidium, that is a suggestions also opted by the current number one of the list, the player named "Barcelona". He says: when selling, only get 75% of the new price. Aging devaluation then comes on top of it.

This would not help in CBV goal gaes though. CBV goal games count ALL the goods/food/pax/mail EVER moved. NCV goals, when for example building a road, loose worth. CBV do not, hence it would not be a solution for CBV goal games. Barcelona however opted this suggestion before CBV goal games hwere introduced ( about a month ago).

No, there is this difference, the measure is valid for any system of valuation,
Because the proposal does not change the value of the vehicle,

If not only the money received when it sells.

The cost was to market the vehicle, move it and put it operating at its destination and therefore is depreciable is part of the book value, but that does not have any value for the new buyer, because the will to repeat all those expenses. Furthermore it is assumed that there is an intermediary who will run their margin of profit. Even considering the rules of the game, the law of supply and demand of depreciating very fast cars, not new, By the rules of the game severely damaging to older vehicles. For all these reasons, my proposal is mainly a proposal to make the game more close to reality.

The linear amortization of the cost of the vehicle has a legitimate rationale, but the money received from the sale of the vehicle need not be identical to its book value, should always be minor, considering the rules of the game, very minor.

[Kurt]

Or just reduce the value of the vehicle on first use (e.g. loading). Then selling the vehicle and buying a new one costs more money than the extra rating cargo and thus making it more expensive to replace the vehicles than not replacing them all the time.

That sounds good. Are there any chances, that this solution is finding a way into the game?
Because i think, that those problems are not only affecting my servers, but the whole openttd-community and gameplay. Everyone, who tries to get the best out of his network is forced to abuse those tricks to get a higher rating.

[PhilSophus]

Or just reduce the value of the vehicle on first use (e.g. loading). Then selling the vehicle and buying a new one costs more money than the extra rating cargo and thus making it more expensive to replace the vehicles than not replacing them all the time.

That sounds good. Are there any chances, that this solution is finding a way into the game?

If the patch you write adheres to the coding style and is of good quality, I don't see why it shouldn't find a way into the game

[Conditional Zenith]

Or you can convince someone else to write it.

[Kurt]

-25% value (on train-engines, vehicles, ships and aircraft) after seven days:

Code: Select all

Index: src/vehicle.cpp
===================================================================
--- src/vehicle.cpp     (revision 14937)
+++ src/vehicle.cpp     (working copy)
@@ -891,6 +891,7 @@
 
 void DecreaseVehicleValue(Vehicle *v)
 {
+       if(v->age == 7) v->value -= v->value >> 2;
        v->value -= v->value >> 8;
        InvalidateWindow(WC_VEHICLE_DETAILS, v->index);
 }

I know, that this is not much, but it's more than nothing.

About station-ratings:

Code: Select all

Index: src/station_cmd.cpp
===================================================================
--- src/station_cmd.cpp (revision 14937)
+++ src/station_cmd.cpp (working copy)
@@ -3047,7 +3048,7 @@
        assert(best_rating1 != 0 || best_rating2 != 0);
 
        /* the 2nd highest one gets a penalty */
-       best_rating2 >>= 1;
+       //best_rating2 >>= 1;
 
        /* amount given to station 1 */
        uint t = (best_rating1 * (amount + 1)) / (best_rating1 + best_rating2);

The penalty seems to make it worse in my opinion. The problem is, that when two have the same rating (or nearly the same), the rating of the second (which is "random", when both have the same rating) is divided by 2.

But that's just one line again. In my opinion, it would be better, when the algorithm gets a rewrite. I will do this tonight, if i have time. The question is: Will the developers accept a new algorithm?


Update:
This diff (of Rev: 14936, Last Changed Date: 2009-01-09 16:11:35) is applying the -25% engine value after seven days and adding a new algorithm for the goods-division. New: Up to four station are able to get stuff. I've tried to keep as much to the original algorithm as possible and tried to not increase the amount of given out stuff too much. I've tested it a bit, but all in all it's very experimental. I would like to hear some comments from the openttd-developers, if this is going into the right direction, or if you dislike it at all. Thank you.
http://stb.gotdns.org/ottd/Other/kurt.diff

[Conditional Zenith]

Older than 7 days isn't quite the same as having been used, but it is easier to code.

And if you are going to remove code, the patch should remove it, rather than comment it out. During balancing it should be fine to leave it, just make sure it is out of the final version (ie. don't forget, I probably would).

As for removing the penalty for being second best, maybe that should only be done if the ratings are sufficiently close to one another?

[Kurt]

Older than 7 days isn't quite the same as having been used, but it is easier to code.

Yes and it doesn't need more computer-resources. It keeps the code simple, but has the same effect: When you buy it by a mistake, you can sell it again without having a big loss. If you use it (>seven days, everything else is not really using), it loses 25%.

And if you are going to remove code, the patch should remove it, rather than comment it out. During balancing it should be fine to leave it, just make sure it is out of the final version (ie. don't forget, I probably would).

Sure
Anything i wrote up to this moment is just testing. I want to know your (/the developers) thoughts about it. When you like the main ideas behind it, i will give my best to produce a fine and high quality patch for you.

As for removing the penalty for being second best, maybe that should only be done if the ratings are sufficiently close to one another?

If the ratings are not close to each other, the division of cargo is anyway more one-sided, so the penalty is (in my opinion) also not needed. But the main balancing will take more tests.

I'm also aiming for the option, to have more than two stations getting cargo. This is already implemented in the diff posted above, but still needs some fine tuning.
Another feature i want to achieve is a better rating, when you provide first resources to a factory. So that you get more of "your" goods, when competing to another player.
The next task is the rating-improvements of new vehicles. I want to reduce the effect a bit and widen the steps (3, 5, 10 years). That's again just a few lines of code and can be done in 5min.

[Conditional Zenith]

Well just to be clear, I'm not a dev (the devs have orange names).

Another thing I just thought of, with your changes, what is to stop someone from setting up lots of stations at a single industry just to get a bigger share of the cargo? I guess that's why the current algorithm works the way it does. Anyway, you should consider this when making your algorithm (maybe only use the best station that a player has).

[belugas]

Well just to be clear, I'm not a dev (the devs have orange names).

Hu?? Damned... I've lost my title :S
orange "only" means the user is a moderator of OpenTTD's section of the forums.

[Conditional Zenith]

I really should pay more attention. Well the devs do have "OpenTTD Developer" under their names. Although you'll probably now point to someone who isn't a dev anymore who still has that title, or someone who is a dev but hasn't got a title yet. Let's just say I'm not a dev.

[Kurt]

Another thing I just thought of, with your changes, what is to stop someone from setting up lots of stations at a single industry just to get a bigger share of the cargo? I guess that's why the current algorithm works the way it does. Anyway, you should consider this when making your algorithm (maybe only use the best station that a player has).

Thanks for that great idea. I will try to include this one.

Last night another idea crossed my mind: I think, it's much better to replace the whole "new train gets higher rating" by "new company gets higher rating". On this way, the game doesn't get harder and new companies have a little advantage over already established ones.
Actually i'm on the move and tomorrow i have not much time, but i will try my best to release another diff as soon as possible.

[zhargon]

Just a quick twopennorth

1) Why do any vehicles have to have a 'saleable' value? - surely it would be much easier to get rid of the 'selling' option altogether, thereby putting the accent on transport management rather than on accounting aberrations. & (I'd have thought), easier to implement

2) I have never seen why anything other than the 'recent history' ((last 24 months?)) of the loading of the cargo should affect the allocation of a specific cargo from an industry. - ie the act of vehicle loading could be the sole determinent for the percentage share of available cargo, divided amongst all stations in range of the industry. Why shouldn't the companies that provide more transport, have an expectation of a greater share? Simpler maths & implementation again. - Getting squashed in nice shiny new carriages may appeal to passengers - but coal etc probably doesn't have any preference.